Situational Influences on Consumers’ Willingness to Pay:
Mood, Stress, and Certainty
Inauguraldissertation
to attain the academic degree
doctor rerum politicarum
(Doktor der Wirtschaftswissenschaften)
at the
ESCP Europe Business School Berlin
by
M.Phil. Erik Maier
Berlin
2013
Doctoral examination committee Head: Prof. Dr. Markus Bick Examiner: Prof. Dr. Robert Wilken Examiner: Prof. Dr. Dr. Helmut Schneider Day of disputation: September 30, 2013
i
Table of Contents
List of figures .................................................................................................................................. ii
List of tables ................................................................................................................................... iii
List of abbreviations ........................................................................................................................ iv
I. Preamble ....................................................................................................................................... 1
2. Situational variables ................................................................................................................. 5
2.1 Emotional reactions: Mood and stress ................................................................................ 5
2.2 Cognitive reactions: Certainty ............................................................................................ 7
II. Manuscripts ............................................................................................................................... 13
1. In the mood to buy? Understanding the interplay of mood regulation and congruence in an
international context ................................................................................................................... 14
2. The impact of stress and level of construal on willingness to pay ......................................... 17
3. The double benefits of consumer certainty: combining risk and range effects ...................... 18
III. Additional empirical investigations ......................................................................................... 19
1.1 Similar consumer collectivism levels ............................................................................... 20
1.2 Differing levels of consumer collectivism ........................................................................ 22
1.3 Summary and limitations .................................................................................................. 23
2.1 Theoretical background: Coping with stress ..................................................................... 26
2.2 Empirical investigation ..................................................................................................... 28
2.3 Summary and limitations .................................................................................................. 29
3.1 Empirical investigation ..................................................................................................... 31
3.2 Summary ........................................................................................................................... 33
IV. Conclusion ............................................................................................................................... 34
2.1 Research implications ....................................................................................................... 39
2.1 Managerial implications .................................................................................................... 42
References ...................................................................................................................................... 49
ii
List of figures
Figure 1: An extension of Russell's (1980) Circumplex Model of Affect ..................................... 10
Figure 2: Summary of findings regarding the impact of consumers’ bad moods on their WTP for
chocolate ......................................................................................................................................... 24
Figure 3: Potential moderating effect of product type on consumers' WTP distributions ............. 27
iii
List of tables
Table 1: Overview of manuscripts ................................................................................................. 11
Table 2: Summary of manuscript results ........................................................................................ 36
Table 3: Summary of additional empirical investigation results .................................................... 37
file:///C:/Users/Erik%20Maier/Documents/Dropbox/Neuer%20Ordner/Updataes%20to%20Diss/Drafts/Manuscript_Situational%20Influences_20130917.docx%23_Toc367207357
iv
List of abbreviations
ANOVA analysis of variance
ed. editor
eds. editors
et al. et alii (and others)
e.g. exempli gratia (for example)
i.e. id est (that means)
€ Euro (currency)
LS less stressed
MS more stressed
NSP no stress-reducing potential
p. page
PANAS positive affect, negative affect scale (Watson et al., 1988)
S-O-R stimulus–organism–response (research paradigm based on Woodworth, 1918)
SP stress-reducing potential
WTP willingness to pay
1
I. Preamble
2
1. Introduction
“In our modern world we have seen inaugurated the reign of a dull bourgeois rationalism,
which finds some inadequate reason for all things in heaven and earth and makes a god of
its own infallibility.”
—John Buchan, A Lodge in the Wilderness (1906, p. 69)
The conservative Lady Amysfort in John Buchan’s book challenged the reign of rationalism
some 50 years before the first research mention of bounded rationality (Simon, 1955). Her asser-
tion that some “inadequate reason” had to be found to retain consistency between the empirical
reality and rationality thus appeared highly prophetic 50 years later. Today, the myth of the fully
rational consumer, with stable preferences and permanent utility maximization, has been ques-
tioned by researchers from many disciplines (Gigerenzer & Selten, 2002). The stance that ration-
ality is limited is standard in marketing textbooks (e.g., Hoyer et al., 2012; Solomon, 2013), even
leading to popular acclaim (Ariely, 2009; Kahneman, 2012). This development has stemmed
from not only psychologists questioning the dogma of rationality (Tversky & Kahneman, 1974)
but also economists challenging rational tenets (Thaler, 1980), similar to conservative Lady
Amysfort’s objection to “bourgeois rationalism”.
As consumer rationality increasingly topples, faced with the dominance of empirically
demonstrated irrational consumer behavior, a group of variables has become increasingly promi-
nent: situational influences (Kakkan & Lutz, 1981). Solomon (2013, p. 359) acknowledges that
“many factors at the time of purchase dramatically influence the consumer’s decision-making
process,” such as salient needs, perceived risks, and emotions (Hoyer et al., 2012). Among these
factors, situational influences are defined as “all those factors particular to a time and place of
observation which do not follow from a knowledge of personal (intra-individual) and stimulus
(choice alternative) attributes, and which have a demonstrable and systematic effect on current
behavior” (Belk, 1974, p. 157). Although this definition has attracted criticism for its limits to a
fixed point in time and negative approach (Barker & Wicker, 1975), as well as its lack of parsi-
mony (Russell & Mehrabian, 1976), its broad nature is what renders it so compelling for the pre-
sent analysis. A list of positive situational variables would always be incomplete; restricting the
definition (e.g., to emotional reactions, Russell & Mehrabian, 1976) would exclude potentially
relevant non-emotional consequences of a situation (e.g., consumer uncertainty).
3
According to Belk’s (1974) definition, the specific characteristics of a situation vary and
function as a stimulus that changes consumer behavior. This situational stimulus elicits a psycho-
logical reaction, which drives the consumer’s behavior (Mehrabian & Russell, 1974). This thesis
applies a classic stimulus–organism–response (SOR) function (Woodworth, 1918) to manipula-
tions of situational variables as the stimuli and thereby aims to explain consumers’ behavioral
responses, in the form of emotional or cognitive changes in the consumer, or organism. The fo-
cus thus is on those situational triggers that are likely to stimulate psychological reactions in the
consumer, such as “momentary moods” or other “momentary conditions” (Belk, 1975, p. 159),
rather than on physical or social surroundings. A consumer’s social surroundings are integrated to
the extent that they interact with situational antecedents though (e.g., the impact of a mood stimu-
lus on behavior could depend on the surrounding group). This thesis analyzes three influences on
consumer behavior: consumers’ mood (e.g., Andrade, 2005), their stress levels (e.g., Pham et al.,
2011), and their certainty about their preferences or product performance (e.g., Wang et al.,
2007).
These situational variables may function as either the subject (i.e., unconscious driver) or
the object (i.e., being consciously driven) of consumer behavior. On the one hand, consumer be-
havior might be unconsciously influenced by various situational variables (e.g., Bargh, 2002;
Dijksterhuis et al., 2005). On the other hand, consumers might consume consciously, in reaction
to the situation or group setting (e.g., Andrade, 2005; Park et al., 1986; Richins, 1994). The focus
of the present investigation is the unconscious drivers of consumer behavior, with consideration
of situations in which conscious decisions overrule unconscious influences (e.g., mood regula-
tion, coping with stress).
Willingness to pay (WTP) provides the dependent variable, for three main reasons. First,
WTP relates directly to a firm’s profitability, through its effect on pricing decisions (Han et al.,
2001). Second, WTP can be measured, to reveal unbiased consumer preferences (Voelckner,
2006). Third, the WTP construct has been conceptually and methodologically extended, with the
introduction of WTP as a range (Dost & Wilken, 2012; Wang et al., 2007). This enhancement
offers a rich means to analyze consumer behavior, both theoretically and empirically.
Following a discussion of the dimensionality of the independent situational variables
(mood, stress, certainty), three chapters each address the impact of a particular variable on con-
4
sumer behavior. Additional empirical evidence supplements these findings. Finally, the last chap-
ter summarizes all the results and offers general implications and routes for further research.
5
2. Situational variables
This thesis considers two realms of situational factors, both of which likely prompt consumer
behavior responses, but based on different mediating psychological reactions. The first realm
refers to situational variables that primarily prompt an emotional reaction to the situational stimu-
lus. Specifically, this investigation considers the role of mood and stress. The second arena per-
tains to consumers’ cognitive reactions and behavioral responses to situations trigger different
degrees of consumer certainty.
2.1 Emotional reactions: Mood and stress
The broadest conceptualization of emotional reactions to a situation consists of three dimensions:
pleasure, arousal, and dominance (PAD, Mehrabian & Russell, 1974). These dimensions appear
orthogonal; Mehrabian and Russell (1974) develop semantic differentials to highlight their bipo-
lar natures (e.g., pleasure uses happy/unhappy; arousal relies on stimulated/relaxed; dominance is
controlling/controlled). Their work contrasts with a prior understanding, namely, that consumer
arousal constituted an antecedent to all emotions (Schachter & Singer, 1962). This assumption
has been empirically discredited by research that shows arousal is an essential component of most
emotional reactions (Bagozzi et al., 1999).
Empirical research also affirms the relevance of two of Mehrabian and Russell’s (1974)
three dimensions: pleasure and arousal. Russell (1980) introduced the Circumplex Model of Af-
fect, which used a factor analytical design to substantiate their conceptualization. Different emo-
tional reactions were summarized as “affect” and positioned in a two-dimensional system, with
differing degrees of pleasure and arousal. Some reactions denoted axes of the system (misery vs.
pleasure; sleepiness vs. arousal); others combined the axes (e.g., distress = misery + arousal).
Despite some general criticism (Richins, 1997), this two-dimensional system has found wide-
spread empirical support (Havlena & Holbrook, 1986; Mano, 1991; Mayer et al., 1991), with
subsequent research adding nuance to the notion (Russell & Barrett, 1999). Consensus affirms the
bipolar nature of each dimension (Cohen et al., 2008), and similar models offer different termi-
nology but similar content (e.g., Watson and Tellegen [1985] refer to arousal as engage-
ment/disengagement).
Other emotional conceptualizations have received less empirical support and remain
somewhat difficult to apply, because they do not relate the different emotional reactions to one
6
another. Plutchik (1980) introduces eight emotional dimensions (joy/sadness, disgust/acceptance,
anger/fear, expectancy/surprise), partially similar to Russell’s (1980) model, but without empiri-
cal support (Havlena & Holbrook, 1986; Plutchik & Kellerman, 1980). The PANAS scale, devel-
oped by Watson and colleagues (1988), distinguishes only between positive (PA) and negative
affect (NA), so it supports the classification but not the relation of various emotional reactions
(e.g., stress and bad moods are both negative affect). Finally, Laros and Steenkamp (2005) struc-
turally relate different emotional reactions but again only through positive–negative distinctions.
Therefore, this thesis relies on the differentiation between pleasure and arousal to distinguish
different emotional reactions.
Mood. In empirical investigations, differences in mood usually are described on a continuum,
according to the pleasure dimension of Russell’s (1980) model. Consensus exists that mood is
bipolar (Cohen et al., 2008), and most mood scales use semantic differentials, such as pleas-
ure/misery (e.g., Swinyard, 1993). Substantial empirical research thus employs the pleasure di-
mension to generate different mood states (e.g., Lopéz Lopéz & Ruiz de Maya, 2012).
However, some authors suggest a second mood dimension: mood management (e.g.,
Mayer et al., 1991; Thayer et al., 1994). Mood management refers to the person’s assessment of
her or his own mood and regulation toward a personally optimal level. Although the relationship
between the perception and management of mood has not been formalized, research on mood
regulation has built on this dimension (conceptually: Andrade & Cohen, 2007; empirically:
Gould, 1997). This thesis applies a mood management perspective where theoretically necessary.
Specifically, beliefs in the changeability, or “transience” (Labroo & Mukhopadhyay, 2009), of
mood serve as potential moderators of mood-regulating behavior, such that a person seeks to reg-
ulate mood only when he or she believes that mood can be changed through such behavior.
Stress. Stress is usually characterized as the opposite of relaxation (Russell & Barrett, 1999). Alt-
hough accompanied by discomfort, stress mainly induces higher levels of arousal, associated with
higher heart rates, increased metabolism, and breathing (Benson, 2001). Arousal thus must be
distinguished from activation: Arousal measures physical reactions to stress, whereas activation
includes mental effort (Gardner, 1986) and is not purely emotional.
The use of the arousal dimension to differentiate stress from relaxation is common. Extant
research has applied this approach (Park et al., 1989; Paulhus & Lim, 1994; Pham et al., 2011).
7
Scales to measure stress and relaxation generally use semantic differentials associated with the
arousal dimension of Russell’s model (e.g., Gorn et al., 1997). Other scales ask participants to
directly state their stress levels in a given situation (e.g., “The situation is not at all/extremely
stressful,” Oliver et al., 2000; “My life was very stressful,” Lee et al., 2007a). Other emotional
components of stress (relaxation) also have been acknowledged, including its negative (positive)
perception. Benson (2001) highlights a close relationship between good mood and relaxation;
both Gorn and colleagues (1997) and Pham and colleagues (2011) recognize the pleasantness of
an affective state in their research on relaxation. Therefore, this study relies mainly on the arousal
dimension to differentiate between stress and relaxation but also acknowledges the unpleasant
and pleasant emotions associated with both states. Positive stress, or eustress (“eu-” is the Greek
root for “good”; Seyle, 1974; see also Le Fevre et al., 2006), is not addressed here.
Current textbooks acknowledge the roles of mood and stress and their dimensionality but
mostly focus on the impact of mood on consumer behavior, without integrating the influence of
arousal (Hoyer et al., 2012; Solomon et al., 2002). This observation suggests the need to establish
stress more strongly as a driver of consumer behavior.
2.2 Cognitive reactions: Certainty
Certainty is a relatively cognitive construct. Although in a specific situation, a consumer might
“feel” very certain or uncertain about her or his own preferences or the future performance of a
product, this state likely has resulted from an active, cognitive evaluation of available consump-
tion options. High attribute conflict could cause consumers to become less certain about their
preferences. For example, a consumer faced with two conflicting dinner choices (healthier and
less tasty versus unhealthy and tasty) probably does not decide intuitively which option he or she
prefers but actively assesses both options’ (dis)advantages. The degree of consumer certainty also
might interact with the situation: If this diner has just finished a long run, he or she might feel
more certain about wanting the less healthy option, because indulging would generate less re-
grets. Again, the consumer decision likely rests on a cognitive evaluation of the person’s current
state, in the current situation.
Various types of certainty have been noted, but two appear most relevant for consumer
behavior: preference and performance (Wang et al., 2007). In line with extant research (Byzalov
& Shachar, 2004; Fischer et al., 2000), the present investigation analyzes the impact of both cer-
8
tainty types on consumers’ WTP. Preference certainty describes the degree to which a consum-
er’s goals are perceived to conflict (e.g., consumer health and taste goals in the dining decision;
Fischer et al., 2000). Performance uncertainty captures the risk of product non-performance (e.g.,
the dinner option will not taste as expected; Rust et al., 1999). Both preference and performance
certainty reflect differences in individual consumers’ evaluations of their consumption options in
a specific situation. Other certainty classifications instead capture certainty about market compo-
sition or development, such as price (Mazumdar & Jun, 1992), knowledge, or choice
(un)certainty (Urbany et al., 1989). Certainty also has been associated, albeit infrequently, with
the pleasure/displeasure dimension (Goldsmith & Amir, 2010; Lee & Qiu, 2009). Because it ap-
pears related mostly to certainty about gifts or prices, this association does not appear in the cur-
rent study.
Furthermore, consumer certainty constitutes a situational stimulus, because most situa-
tions drive consumer certainty perceptions, creating intraperson variance in consumer behavior.
Although preference certainty is more intuitively situational, performance certainty also is strong-
ly influenced by situational variables. The conflicting dinner options scenario highlights the high-
ly situational nature of preference certainty. First, the options available across multiple situations
might affect preferences. The consumer faces some situations in which he or she would have to
evaluate both consumption options separately and not jointly, whereas in other situations, the two
options combine with other options. A low degree of preference certainty arises from the utility
tradeoff of the two options (less tasty and healthier vs. tastier and less healthy). Had the evalua-
tion taken place separately or combined with a less contrasting or inferior option (e.g., less tasty
and not healthy), the degree of preference certainty would have been higher. Second, the situation
might alter the weights of the different option attributes, creating more or less conflict and thus
more or less certainty about preferences (i.e., a long jog might make the decision for a less
healthy option easier). Third, preferences are mainly constructed during the preference elicitation
process (i.e., in determining own preferences; Gregory et al., 1993), so they are prone to situa-
tional influences, such as additional information or the consumer’s previous actions.
Performance certainty, in contrast, is not intuitively situational, as the expected perfor-
mance of a product or service should be stable. However, the expected performance of a product
is not objectively measurable or constant, because consumers do not possess complete infor-
9
mation about the product. Therefore, consumers must estimate its future performance, and this
estimation depends strongly on the situation, as four examples highlight:
1. The consumer may possess more or less information that is relevant to estimations of fu-
ture product performance (e.g., information on a washing machine’s number of options,
energy efficiency, or objective test results might be available during the decision process,
or not).
2. Feedback from the market could influence the evaluation situationally (e.g., current avail-
ability, valence, and variance of product reviews for washing machines, as well as previ-
ously received word of mouth; Heiner, 1985).
3. The utility payoffs might be unstable, depending on the situation (e.g., product becomes
outdated with the introduction of a new technology, such as combined washing machines
and dryers).
4. Performance expectations might rest on past experience with the product or product cate-
gory (e.g., previous acquisitions, product testing at the retailer). The presence of such ex-
periences is also situational, because a consumer, for instance, might (not) have the op-
portunity to test the product while shopping.
Thus, though certainty about a product’s performance should be intra-personally stable, in reality
it is strongly influenced by the decision situation.
Finally, research on WTP as a range notes the presence of within-consumer variance in
the WTP range (e.g., Dost & Wilken, 2012; Wang et al., 2007). Therefore, situational and non-
individual factors must exist, as well as drive consumer behavior.
For this study, certainty represents the opposite of uncertainty. The described research
differentiates different degrees of certainty, though extant research mainly discusses the effects of
uncertainty. This terminological distinction does not constitute a conceptual difference; certainty
and uncertainty locate at the opposite ends of a single continuum.
10
3. Introduction: Manuscripts and empirical investigations
The subsequent chapters each deal with one of the focal situational variables: mood, stress, and
certainty. Figure 1 shows their conceptual relation. Mood and stress rely on Russell’s (1980)
model of affect; certainty is independent. Mood is operationalized by the pleasure/displeasure
dimension, and stress uses high and low levels of arousal. The operationalizations of both varia-
bles thus is orthogonal, though interactions exist (see Section 2.1). Certainty constitutes an addi-
tional dimension.
Figure 1: An extension of Russell's (1980) Circumplex Model of Affect
Although these chapters are linked through their focus the behavioral impact of situational
variables, each manuscript constitutes an independent contribution to research, with distinct rele-
vance for their respective fields. Table 1 provides an overview of the manuscripts, their contribu-
tions, and publication status.
The first contribution, “In the Mood to Buy? Understanding the Interplay of Mood Regu-
lation and Congruence in an International Context,” introduces the moderating role of culture to
the mood–WTP relationship. Specifically, four studies show that consumers’ tendency to use
mood regulation when in a bad moods and faced with a mood-lifting product depends on their
degree of collectivism. Individualist consumers (from Germany) refrained from mood regulation
and consistently behaved mood congruently as their WTP levels decreased. Collectivist consum-
ers (from Turkey) instead used consumption to regulate their negative mood when the product in
question offered the chance, which increased their WTP. These behavioral differences can be
11
explained by higher mood transience beliefs about the extent which one’s own mood can be
changed in collectivist versus individualist cultures.
Table 1: Overview of manuscripts
The second manuscript, “The Impact of Stress and Level of Construal on Willingness to
Pay,” establishes the negative impact of stress on consumer valuations of products. This effect
rests on changes in consumers’ construal levels, associated with stress (low construal) or relaxa-
tion (high construal), which unconsciously influences importance perceptions and consumers’
valuations. However, the negative effects of stress can be mitigated if the product’s level of con-
strual matches the consumer’s.
Finally, “The Double Benefits of Consumer Certainty: Combining Risk and Range Ef-
fects” investigates the link between consumer certainty and WTP ranges. The results demonstrate
empirically the conceptually established negative effect of certainty on WTP range size. The
analysis also suggests that certainty increases consumers’ expected WTP, irrespective of their
position, through a stronger increase of consumers’ floor relative to ceiling prices.
Additional empirical investigations in Chapter III replicate findings about the moderating
role of culture in the mood–WTP link in a broader European context (Section 1), show the mod-
erating role of products’ stress-reducing potential on the relationship between stress and WTP
12
(Section 2) and aim to establish review variance as a means to manipulate consumer certainty and
their WTP range (Section 3).
13
II. Manuscripts
14
1. In the mood to buy? Understanding the interplay of mood
regulation and congruence in an international context
Manuscript No. 1
This manuscript was published as: Maier, E., Wilken, R., Schneider, H., & Kelemci Schneider,
G. (2012). In the mood to buy? Understanding the interplay of mood regulation and congruence
in an international context. Marketing Letters, 23(4), 1005–1018.
DOI: 10.1007/s11002-012-9200-7
15
Addition to the manuscript: limitations not presented in the publication
We limited our research in manuscript 1 purposefully to a specific experimental design; thus,
several opportunities for further research remain. Generally, the experimental character of Studies
1, 3, and 4 may be a concern. Although laboratory settings help isolate the effects of interest,
their applicability to acquisitions at the point of sale may be limited, despite our application of
incentive-compatible WTP measures in Study 4. We analyzed the impact of good or bad moods
on only two sample products, which consumers do (chocolate) or do not (sponges) associate with
mood-lifting capabilities. Although we validated these manipulations, we cannot guarantee that
consumers might not react differently to other mood manipulations or different (mood-lifting)
products. Our operationalization of mood transience on the basis of the different sample countries
also could constitute a limitation, in that international surveys are susceptible to several biases
(Tellis & Chandrasekaran, 2010).
Further research should tackle these limitations. For example, the impact of mood on
WTP might be tested in a retailing environment. Additional analyses could address the impact of
other mood states (e.g., neutral) or mood orientations (e.g., avoidance) on consumer behavior.
With regard to product choice, the dominance of mood-congruent versus -regulating behavior
could be assessed for “mood-threatening” products too. Andrade (2005) has analyzed the impact
of mood on the choice of negatively perceived options, though he used nonmonetary evaluations
as a dependent variable and did not assess the impact of beliefs about mood transience as mod-
erator.
Our understanding of the moderating role of mood transience beliefs might be extended
too. The mood–WTP link should be analyzed in other countries. Research thus could replicate
our findings in countries that diverge equally in terms of collectivism/individualism but that are
more geographically distant. In turn, contrasting our findings for highly collectivistic versus high-
ly individualistic cultures with countries characterized by lower levels could extend our results.
Research could also determine if variables other than product-related traits and beliefs
about mood transience (culture) drive the impact of mood on WTP. Personality traits such as ex-
traversion or neuroticism influence evaluations of stimuli (Mooradian, 1996). Vivid memories of
past regulation experiences could increase beliefs in the success of mood regulation (Luomala,
2002; Pollai et al., 2010). Furthermore, Pham et al. (2011) find that the degree of relaxation (vs.
stress) increases monetary evaluations by consumers in good moods. Particularly considering the
16
mood-lifting capabilities of certain products, analyzing a stress scenario with participants in bad
moods could generate further insights. For example, stressed consumers might try to regulate
their stress levels by consuming relaxing products.
Finally, we based the mood–WTP link on mood transience, which depended on consum-
ers’ collectivism. Cross-cultural studies on links between consumers’ feelings (other than mood)
and behavioral measures (other than WTP) should consider reliance on feelings or associated
constructs (need for affect, need for emotion, need for cognition, or faith in intuition). These con-
structs also vary with collectivism, and they may apply, beyond transience beliefs, as a rationale
for such links.
17
2. The impact of stress and level of construal on willingness
to pay
Manuscript No. 2
Authors: Erik Maier and Robert Wilken
Publication status: Under review in Psychology & Marketing.
Manuscript available from the author upon request.
18
3. The double benefits of consumer certainty: combining risk
and range effects
Manuscript No. 3
Authors: Erik Maier, Robert Wilken and Florian Dost
Publication status: Revise and resubmit in Marketing Letters.
Manuscript available from the author upon request.
19
III. Additional empirical investigations
20
1. Impacts of mood on WTP in European contexts
Manuscript 1 suggests that one way of segmenting European markets might be according to their
individualism, due to the moderating role of culture in the relationship between mood and WTP.
This finding contradicts intuitive and preconceived notions of segmentation in Western Europe,
where a country’s location (e.g., South vs. North) and cultural peculiarities (e.g., extroverted Ital-
ians vs. reserved Finns) generally has been used to predict different consumer behaviors. For ex-
ample, a survey conducted on the 50th anniversary of the Elysée Treaty between France and
Germany revealed that French people associate diligence and parsimony with Germans, while
Germans associate indulgence and savoir vivre with their French neighbors (Die Zeit, 2013). A
well-established segmentation of Western Europe thus distinguishes the Northern or Germanic
segment from the Mediterranean or Latin segment (Carr & Texeraud, 1993; Gatley et al., 1996).
This distinction has resonated in both scientific literature (e.g., Jesuino, 2012) and marketing
practice (e.g., Cova & Halliburton, 1993; Kaynak & Jallat, 2005). Whereas consumers from the
Mediterranean region are usually characterized as “suave, refined, rich and enjoy[ing] hedonic
pleasure” (Kabiraj & Shanmugan, 2011, p. 295), their northern counterparts are said to possess a
“tradition of orderliness, standards, and rules” (Chandrasekaran & Tellis, 2008, p. 853).
This empirical investigation aims to replicate the findings from Manuscript 1 regarding
the impact of culture on the mood–WTP link in an extended European setting of four countries
(France, Germany, Greece, and Italy). Thus it might add validity to the presented challenge to the
segmentation of European markets into simply North/Germanic versus Mediterranean/Latin. The
first empirical investigation contrasts France and Germany directly; these two countries exhibit
similarly low collectivism scores but typically are associated with different clusters (Mediterra-
nean vs. Northern). A second study extends this analysis to more countries with varying degrees
of collectivism (Greece and Italy compared with Germany), to gain finer-grained insights into the
effect of mood on WTP and determine if traditional segmentation approaches should be modi-
fied.
1.1 Similar consumer collectivism levels
Method. Two countries that do not differ in their individualism but are usually clustered into dif-
ferent country segments were selected: France as part of the Mediterranean cluster and Germany
as part of the Northern cluster (House et al., 2004; Ronen & Shenkar, 1985). Both countries score
21
equally high on individualism (France = 71, or 9th of 66 countries; Germany = 67 or 14th; Hof-
stede, 2001) and low on societal in-group collectivist practices (France = 4.37 or 49th of 62 coun-
tries, Germany = 4.02 or 56th; House et al., 2004). A pretest also confirmed the two countries’
cultural similarity in terms of their collectivism (MFrance = 3.98, MGermany = 3.89, F(1, 133) = .3, p
= .58; House et al., 2004). They did not differ in potential confound variables (e.g., Gaston-
Breton & Martin, 2011; Lin et al., 2006), such as materialism (MFrance = 4.20, MGermany = 4.15,
F(1, 133) = .06, p = .81; Richins & Dawson, 1992) or openness to feelings (MFrance = 3.81, MGer-
many = 3.74, F(1, 133) = .53, p = .47; Costa & McCrae, 1985).
As in Manuscript 1, chocolate served as the mood-lifting product. A pretest supported the
appropriateness of this selection: Chocolate was consistently perceived as mood lifting (“Every
now and then I buy chocolate to improve my mood,” test for difference from scale mean of 3.5,
M = 3.98, t = 1.84, p = .07) and significantly more hedonic than utilitarian on scales proposed by
Khan and Dhar (2010) (test for difference from scale mean, M = 2.17, t = -7.91, p < .001) and by
Voss and colleagues (2003) (paired t-test, MHED = 4.75,. MUT = 3.74, t = 4.1, p < .001). Percep-
tions of mood-lifting and hedonic characteristics did not differ for Germany and France.
Participants from each culture were randomly assigned to either the good or bad mood
condition. The survey was administered online. The procedure, manipulation, manipulation
checks, and the calculation of the WTP index were similar to those for Manuscript 1.
Results. Of the 158 participants who completed the survey, 5 were excluded because they offered
extreme WTPs, more than twice the standard deviation for their respective group. The remaining
153 participants (60 from France, 68% female; 93 from Germany, 49% female) had an average
age of 24.75 years. For the 73 participants in the good mood stimulus and the 80 respondents in
the bad mood stimulus, the manipulations of mood (F(1, 152) = 178.2, p < .001, η = .74) and
product perception (paired t-test, t = 4.94, p < .001) were successful, across both countries.
According to a 2 (mood) × 2 (country) analysis of variance (ANOVA), when they were in
bad moods, people’s WTP decreased in both France (mean decrease = 4.2%; MWTP, good = 1.90 €,
MWTP, bad = 1.82 €) and Germany (mean decrease = 20.8%; MWTP, good = 1.43 €, MWTP, bad = 1.13
€). As expected, the main effect of mood was significant (F(1, 152) = 4.44, p < .05, η = .17),
whereas the interaction effect was not (F(1, 152) = 1.39, p = .25, η = .09).
22
Although this insignificant interaction indicated that consumers from both countries be-
haved similarly, caution is warranted. First, the absolute WTP levels differ considerably, and the
insignificance of the interaction mainly reflects the large standard deviations. Second, though the
p-value of the interaction effect considerably exceeds the 5% threshold, the effect size is nearly
small (Cohen, 1988). Third, the likelihood of a type-II error β (i.e., falsely accepting the null hy-
pothesis) is approximately 80%, considerably greater than the common threshold of four times
the p-value (Cohen, 1988). Although this first investigation thus constitutes a replication of the
findings of Manuscript 1, namely, that individualism moderates the mood–WTP relationship, the
results should be treated with caution and further replicated.
1.2 Differing levels of consumer collectivism
Method. This second investigation aims to substantiate the moderating character of culture in an
extended country setting that includes two additional countries from the Mediterranean cluster
that differed considerably in their individualism. Italy is highly individualistic (76, 6th of 66
countries; Hofstede, 2001) and low on collectivism (4.94, or 41st of 62 countries; House et al.,
2004). In contrast, Greece scores low on individualism (35, or 38th) and high on collectivism
(5.27, or 35th). Both countries are associated with the Mediterranean cluster (Ronen & Shenkar,
1985), though Greece is sometimes grouped with Eastern Europe (House et al., 2004). The com-
parison of these two countries included a different German sample, to provide an established in-
dividualistic benchmark.
Similar to the previous investigation, participants from each culture were randomly as-
signed to the good or bad mood condition for an online survey. Each respondent confronted the
same mood-inducing stories and the same manipulation checks. The WTP index for each country
was calculated as previously described.
Results. The survey was completed by 214 respondents, but 5 were excluded due to their extreme
WTP outliers. The remaining 209 participants (48 from Greece, 46% female; 108 from Italy, 41%
female; 53 from Germany, 58% female) were 26.37 years on average. The experiment induced
101 participants to be in a good mood and 108 in a bad mood. Again, the manipulations of mood
(F(1, 208) = 149.32, p < .001, η = .65) and product perceptions (t = 7.76, p < .001) were signifi-
cant.
23
As expected, participants from the individualist countries lowered their WTP when they
were in bad moods: Italy by 15.6% (MWTP, good = 2.11, MWTP, bad = 1.78) and Germany by 6.0%
(MWTP, good = 1.89, MWTP, bad = 1.77). In contrast, consumers in collectivist Greece paid a premium
(mean increase = 10.6%; MWTP, good = 1.95, MWTP, bad = 2.15). A 2 (mood) × 2 (culture: collectiv-
ism vs. individualism) ANOVA used the pooled data from Germany and Italy as individualist
countries. As predicted, consumer culture significantly moderated the relationship between mood
and WTP (F(2, 208) = 3.1, p < .05, η = .17), but the main effect of mood remained insignificant
(F(2, 208) = .02, p = .89, η = 0, β < .05). This result reaffirms the appropriateness of individual-
ism–collectivism as a segmentation criterion, rather than the traditional North–South divide.
1.3 Summary and limitations
Figure 2 summarizes the results of these empirical investigations into the relationship be-
tween consumers’ mood and their WTP for a mood-lifting product. In line with the predictions
from Manuscript 1, the present studies establish a mood-congruent effect in countries that score
relatively high on individualism (Germany, France, Italy), even though they do not belong to the
same geographical cluster. In contrast, Greek consumers (low individualism) exhibit mood-
regulating behavior, such that consumers in bad moods express a higher WTP than consumers in
good moods. These findings are in line with recent research that has demonstrated mood-
regulating behavior in collectivist Spain (López López & Ruiz Maya, 2012).
Firms dealing with multicultural markets (even in supposedly homogeneous geographical
clusters) must use their knowledge about consumers’ individualism and collectivism, or some
chronic measures, to adapt their advertising, promotions, and displays. The international market-
ing mix should be adjusted to match consumers’ culturally unique beliefs about their ability to
change their moods through consumption. In collectivistic countries such as Greece, emphasizing
the mood-lifting capabilities of a product can increase consumers’ valuation of that product. Con-
sumers in a bad mood likely identify mood-lifting cues and are willing to pay a premium to im-
prove their mood. In individualistic countries such as Italy or France, a similar package and ad-
vertising may be less effective. To improve individualistic consumers’ WTP, store designs and
staff behaviors should seek to ensure that customers experience positive moods at the point of
sale; any deterioration decreases the price they are willing to pay, irrespective of product type.
24
Figure 2: Summary of findings regarding the impact of consumers’ bad moods on their WTP for chocolate
However, the findings from the additional empirical investigation also feature some limi-
tations. First, the results indicate sizable differences in the test results for Germany across studies:
The effect of negative mood in the first investigation decreased consumers’ WTP by 21%, but its
effect in the second investigation was only 6%. The difference could be due to the non–incentive-
aligned form of WTP measurement, which might increase WTP variance, but it raises doubts
about the reliability of the results. Second, whereas the first empirical investigation showed a
similar impact of bad mood in both Germany and France (insignificant interaction effect), the
likelihood of falsely accepting the null hypothesis remained high. Third, the sample composition
in the second investigation was unbalanced (only 25% collectivistic). Fourth, the analyses were
conducted using participants’ passport nationality as the indicator of their culture, as a nominal
independent variable. Studies that assess consumer collectivism on an individual level and con-
duct regression analyses could produce more fine-grained results, circumvent the danger of type-
II errors, and evade issues with unbalanced designs. Fifth, the means and timing of the data col-
lection could have distorted the results. Online surveys may induce strategic answering behavior
(Lusk et al., 2007) and yield the danger of non-random selections (Wells et al., 2012). The data
collection also lasted more than a year: For investigation 2, data from Greece and Germany were
collected in 2010, and in the midst of the economic crisis, but data from Italy were collected in
25
late 2011. This difference in timing could also explain the measured behavioral differences, as
that, for instance, consumers during an economic crisis could have compensated for their person-
al economic difficulties with indulging with hedonic products, rather than aiming to regulate their
mood.
26
2. Stress, coping, and willingness to pay
Although Manuscript 2 detailed the unconscious impact of different degrees of stress on consum-
er cognition and product valuation (i.e., stress as the subject), it ignored consumer behaviors de-
signed to reduce stress levels (i.e., stress as the object). Yet abundant research on coping with
stress (Duhachek, 2008) predicts an influence of coping on consumer behavior. Specifically, cop-
ing potential likely improves the product’s valuation under stress, through additional value gener-
ated from a product’s stress-reducing capabilities. This effect would run counter to the finding of
a negative effect of stress on consumer valuations. This investigation therefore offers an initial
test of the role of coping in consumer valuations.
2.1 Theoretical background: Coping with stress
In contrast with unconscious reactions to stress (Manuscript 2), coping with stress is a conscious
act, based on the perception of a situation as stressful, in which “Appraisal is the cognitive pro-
cess through which an event is evaluated with respect to what is at stake (primary appraisal) and
what coping resources and options are available (secondary appraisal)” (Folkman & Lazarus,
1980, p. 223). The intensity of the coping response depends on the severity of the perceived
stressor (Thoits, 1995). This situational assessment includes an evaluation of the situation, past
experiences, and current beliefs (Duhachek, 2008; Thoits, 1995). As a consequence, consumers
exhibit interindividual (Cooper et al., 1992; Duhachek, 2005) and situational (Mick & Fournier,
1998) differences in their coping behavior.
The most common classification of coping behavior distinguishes between problem- and
emotion-focused coping (Lazarus & Folkman, 1984). Problem-focused coping targets the per-
ceived environmental stressor; emotion-focused coping aims to regulate the consumer’s emotion-
al reaction to the stressor. Although other classifications have been introduced (e.g., Mick &
Fournier, 1998), the traditional distinction between problem- and emotion-focused coping re-
mains valid (Yi & Baumgartner, 2004). This investigation focuses on emotion-focused coping,
because it might be linked to changes in consumer behavior (e.g., acquisition of a stress-relieving
product), whereas problem-focused consumption is a less common initiator of consumption (e.g.,
acquisition of conflict resolution tools, drugs enhancing work capabilities).
Emotion-focused coping is more likely when a consumer experiences a stressful situation
as uncontrollable (Thoits, 1995) or unexpected (Moschis, 2007). Consumers might apply multiple
27
coping strategies in parallel (Duhachek & Kelting, 2009). If a consumer resorts to coping through
consumption (Moschis, 2007), evaluations of products that offer coping potential likely improve,
through anticipation of a positive emotional reaction. However, to be suitable for consumption
coping, a product must offer stress-reducing or relaxation potential (e.g., drinking alcohol,
Cooper at al., 1992). If instead a product offers no coping potential (Figure 3, left side), constru-
al-level effects should drive its evaluation (see Manuscript 2). Consumers’ WTP then might de-
crease under more, relative to less, stressful conditions. Only if the product’s construal level
matches the consumer’s is a mitigation of the negative impact of stress possible (see Manuscript
2, study 3).
Figure 3: Potential moderating effect of product type on consumers' WTP distributions
Even if a product offers coping potential (Figure 3, right side), not all consumers under
stress resort to coping behavior. Some consumers prefer not to apply the product’s coping poten-
tial (bottom right, solid line), and others might vigorously aim to cope, leading to greater WTP
under more stress (bottom right, dotted line). The net effect is unclear; because the effects run
contrary to each other, they likely to increase the WTP variance for consumers under more rela-
tive to less stress. Previous research into the effect of negative moods on consumer valuation of
mood-lifting products, including Manuscript 1, proposes similar effects of the stimulus on WTP
variance. Applied to the regulation of consumers’ stress levels:
H1: For products with stress-reducing potential, consumers exhibit greater WTP variance
when they are under more relative to less stress.
28
Although consumers’ propensity to use coping is unclear, the positive effect of some cop-
ing behavior is likely to increase the average valuation of a product with stress-reducing potential
relative to a product without such capabilities. The WTP net effect for a product with coping po-
tential depends on the magnitude of coping behavior among consumers (Figure 3, bottom right).
The more common is it so cope through consumption, the more strongly the negative effect of
stress on valuation gets mitigated, or even reversed if many consumers resort to coping. Formal-
ly:
H2: The product category moderates the relationship between consumer stress and WTP,
such that the negative effect of more relative to less stress on WTP is mitigated or re-
versed for products with, relative to products without, stress-reducing potential.
2.2 Empirical investigation
Method. Two products that differed in their potential to help consumers cope with stress were
selected: wine (high potential) and sponges (low potential). In a pretest with 13 products and 19
participants, wine was described as the most relaxing product (“The product helps me relax,” 1 =
“Do not agree at all” to 5 = “Fully agree”; M = 3.85; difference from scale mean: t = 3.65, p <
.01) and sponges as the least relaxing (M = 1.00; SD = .00). The products differed significantly in
their relaxation and thus their coping potential (within-sample t-test: n = 19, t = 11.6, p < .001).
The selection was confirmed with a second question item (“The product helps me reduce stress”;
MWine = 3.48, MSponge = 1.00).
The survey was conducted in an experimental setting among graduate students. To ma-
nipulate stress levels, the participants in the experimental group were confronted with a simple
computation task, masked as annual assessment of students’ quantitative capabilities: They were
asked to subtract 17 from 1,194 as often as possible and note the resulting series. This procedure
raises people’s arousal levels and causes stress (Benson, 2001). In a supposedly unrelated study,
participants later stated their WTP and arousal levels (Mehrabian & Russell, 1974). Finally, par-
ticipants had to qualitatively guess the purpose of the survey. The control group received no
stress-inducing treatment. A WTP index for each product was calculated with an expected value
of 0 for the control group under lower stress (i.e., indicating the WTP change under stress).
Results. Of the 46 participants who completed the survey (mean age = 23.2 years; 71% female).
4 were excluded because they guessed the purpose of the survey and 5 because they stated WTP
29
levels that exceeded twice the standard deviation within the product, leaving 37 participants in
the sample. The remaining participants differed marginally significantly in their arousal levels
(MLess stressed = 4.18, MMore stressed = 3.5; F(1, 36) = 3.8, p = .06, η = .31).
In line with H1, participants’ WTP variance for wine in the more stressed conditions ex-
ceeded the WTP variance in the less stressed control condition by four times (F(1, 36) = 5.0, p <
.01). Though not formally hypothesized, Figure 3 implies the similarity of WTP variance for
non–stress-reducing products, in that coping for consumers under stress was unlikely. As ex-
pected, WTP variance for sponges did not differ across groups (F(1, 36) = 1.2, p = .35).
As predicted by H2, more stressed consumers’ WTP decreased significantly for sponges
(MLess stressed = 1.61 €, MMore stressed = 1.14 €; F(1, 36) = 4.1, p = .05, η = .32) and increased margin-
ally significantly for wine (MLess stressed = 4.18 €, MMore stressed = 5.03 €; F(1, 36) = 3.2, p = .08, η =
.28). That is, participants’ WTP showed a significant moderating effect by product category for
more relative to less stress (MWine = +20%, MSponge = −29%; within-sample t = 4.8, p < .001).
2.3 Summary and limitations
These findings provide some evidence of a moderating effect of product category in the stress–
WTP relationship. A significant portion of consumers uses consumption to cope with stressful
situations: If a product has stress-reducing potential, their WTP likely increases when they expe-
rience higher, relative to lower or no, stress. This effect is particularly noteworthy because stress
often is perceived negatively or accompanied by bad moods. Following Manuscripts 1 and 2, the
effect of stress and negative affect should have been unidirectional in the absence of coping be-
havior. The increase in WTP for wine, as a stress-reducing product, suggests that increasing con-
sumer valuation from coping might exceed the overall negative evaluation (Manuscript 2) or
counter potential interference from negative moods (Manuscript 1).
Therefore, marketing practitioners might exploit coping behavior to increase consumers’
WTP for a product when they are under stress. The marketing mix could be adjusted to position a
product as stress-reducing. If many consumers are likely to be stressed (e.g., during rush hours),
their average WTP could be increased by adjusting in-store advertising to promote the stress-
relieving nature of these products. Furthermore, there is no indication that consumers who re-
laxed then perceived stress-reducing products any more negatively. Thus, even if the product
description cannot be altered according to expected consumer stress levels (e.g., by the producer
30
of a product, which cannot alter the package design), a referral to its stress-relieving nature would
not deteriorate product perceptions for consumers suffering from less stress. A retailer that sells
products commonly perceived as stress relieving (e.g., wine merchant, candy store) also might
increase WTP by creating stress for consumers or, as a less drastic means, convincing them that
they are in a stressful situation that must be countered (e.g., salespeople noting the many daily
hassles people face and how a glass of wine might help).
The applicability of these recommendations is limited in three ways though. First, the evi-
dence came from a small sample. Although the described effects were similar in non-parametric
tests (Mann-Whitney U-test), further analyses in different settings and with larger samples should
seek to substantiate these results. Second, most participants were recruited in a university context
and participated only in a hypothetical, non–incentive-aligned survey, which limits the validity of
the findings. Third, the investigation featured only two sample products that contrasted very
strongly in terms of consumers’ perceptions of their stress-reducing potential. To generalize the
results, further research should include more products in the analysis and investigate the effects
of stress with a larger sample, potentially at the point of purchase and with non-hypothetical
WTP measures.
31
3. Impacts of review distributions on consumer certainty and
WTP as a range
This investigation aims to replicate previous findings from Manuscript 3 with an analysis
of the influence of different review distributions on consumer certainty and WTP ranges. Manu-
script 3 established reviews as one means to increase consumer certainty (study 1) but only by
differing the number of reviews (e.g., Jang et al., 2012). The present study employs differences in
the review distribution to manipulate consumer certainty levels. Prior investigations adopt differ-
ent review distributions to manipulate the polarization of consumer reviews (Shao, 2012). Thus,
decreasing review variance (i.e., less polarized responses) should be associated with higher con-
sumer certainty. Reviews for two product could yield the same mean (e.g., 3/5 stars), but percep-
tions of product A could be very polarized (e.g., many consumers with either 1-star or 5-star rat-
ings), whereas perceptions of product B might be very similar (e.g., many consumers rate product
around the mean), which determines the level of review variance. The mean product perception
likely reflects the review distribution much more closely for product B (low variance), so con-
sumers should be more certain of this product’s (average) performance than of the performance
of product A (high variance), which could be either very good or very bad.
This certainty manipulation is particularly relevant in online settings (Zhu & Zhang,
2010), where consumers rely heavily on peer-generated reviews for information (Ludwig et al.,
2013; Weathers et al., 2007). It might be difficult to influence the valence of consumer reviews,
but marketing managers could relatively easily increase the number of reviews, which should
generate a certainty effect of its own (see Manuscript 3, study 1) and also cause more normally
distributed reviews, with smaller variance. This investigation therefore tests the effect of review
valence on WTP ranges, with the expectation that decreasing review variance causes higher cer-
tainty and thus, similar to Manuscript 3, (a) increase expected WTP levels, (b) decrease the WTP
range, and (c) increase the floor price more strongly than the ceiling price.
3.1 Empirical investigation
Method. Consumers received quantitative reviews in a 5-star system of different variance (more
certain: Var = .4; less certain: Var = 3.8) but equal valence (M = 3 out of 5 stars) and number (n =
27) for two services associated with high performance uncertainty (hotel accommodation, spa
visit; see also Jang et al., 2012). In the more certain condition, the reviews were normally distrib-
32
uted, with the majority of reviews similar to the mean review (i.e., inverted u-shape); in the less
certain conditions, reviews were strongly polarized at the extremes (i.e., u-shape).
The review perceptions were confirmed in a prestudy: Lesser review variance was associ-
ated with higher degrees of certainty (M = .59) than greater review variance (M = -1.40; index of
three items: “How certain are you that the product will function?” “How well can you judge how
the product would function?” and “I feel the product would probably not work properly/work
properly”; –3 = low certainty, +3 = high certainty; within-sample t-test: n = 28, t = 7.6, p < .001).
The pretest also included a third review distribution of medium variance (Var = 2.2) whose cer-
tainty perceptions were located between the former two manipulations (M = –.7).
The orders of services and review variance presentation were random. The survey was
conducted in an online setting with Amazon MTurk. Before each evaluation, participants were
introduced to the scenario (trip to Paris/New Orleans; looking for accommodation/spa), then
viewed the reviews, along with brief, standardized information about the service (e.g., location of
the hotel, room size) for at least 30 seconds. They next answered questions that identified non-
attentive participants for exclusion, and finally, they estimated their WTP and stated their certain-
ty levels, as manipulation checks.
Results. Of the 198 participants who completed the survey (56% female, mean age = 33 years),
72 had to be excluded due to non-attention to the study’s content, as evaluated by the control
questions (e.g., “Which service were you just introduced to?” hotel accommodation, spa visit,
dinner in a restaurant; “Which sources of information were presented to you about the service?”
online reviews, press clippings, general information), or because their expected WTP exceeded
twice the standard deviation of the respective group. The sample characteristics did not change
substantively due to the sample reduction (50% female, mean age = 33 years).
Participants’ WTP ranges for the hotel accommodation were in line with expectations,
though insignificant: Expected WTP increased by 9% with more (MC) versus less certainty (LC;
MMC = $120.8, MLC = $110.6; F(1, 125) = 2.3, p = .135, η = .13). The WTP range decreased but
only by 3% (MMC = 43.9%, MLC = 44.3%; F(1, 125) = .1, p = .75, η < .1). The floor prices
(+10%) increased slightly more strongly than the ceiling prices (+9%, t = .36, p = .7). The insig-
nificant results matched the main study’s manipulation check, which showed only an insignifi-
cant difference in consumer certainty (MMC = 4.47, MLC = 4.35; F(1, 125) = .22, p = .63, η < .1).
33
In contrast, the stated WTP ranges for the spa visit remained inconclusive and insignificant: With
increasing certainty, the expected WTP increased slightly (-2%, F(1, 126) = .05, p = .8, η < .01),
the range hardly decreased (-7%, F(1, 126) = .6, p = .4, η < .1), and the floor price increased by a
small margin (+2%), while the ceiling price decreased (-4%, t = 1.7, p = .09). This mixed evi-
dence is also reflected in consumer certainty perceptions, which did not differ between the two
manipulations (MMC = 4.07,. MLC = 4.18; F(1, 126) = .23, p = .63, η < .1).
3.2 Summary
The results of this survey were inconclusive. Although the WTP range for the hotel accommoda-
tion developed directionally, as expected, the changes were not significant. The data for the spa
visit remained contradictory (e.g., higher ceiling price under less certainty). However, this finding
does not mean certainty levels cannot be manipulated with review variance. First, the prestudy
confirmed the different consumer certainty levels. Second, the data quality of the MTurk survey
was very low; participants seemingly did not pay close attention to the reviews, as indicated by
the unsuccessful manipulation checks. The impact of review variance on certainty and WTP
ranges thus demands further investigation.
34
IV. Conclusion
35
1. Summary of findings
The combined results of this research imply that situations generally influence consumer behav-
ior. Although this thesis has analyzed the impact of only three exemplary variables (consumers’
mood, stress levels, and certainty about their own preferences or product performance), it offers
notable indications that surrounding situations prompt behavioral responses for other variables as
well. First, the three variables created significant consumer behavior responses in each field of
investigation, as well as across all three areas, in 15 studies. Second, the effects of the situational
variables seemed strong enough to affect consumer behavior, despite the likely presence of other
situational (e.g., weather) or non-situational (e.g., individual product perceptions) variables, par-
ticularly in non-experimental settings. Third, though the focus of this research was on WTP as a
dependent variable, situational variables also affected other behavioral measures (e.g., percep-
tions of feature importance, Manuscript 2).
Consumers’ behavioral responses to situational stimuli appeared unconscious in many
cases though. This finding appears intuitive for emotional reactions to situational stimuli: Con-
sumers should be unaware that their mood state drives their retrieval of mood-congruent infor-
mation (Table 2, part 1). Most consumers are not even generally aware of how their product per-
ceptions evolve (Kahneman, 2012). The finding that consumers are ignorant of the impact of dif-
ferent levels of construal on the way they perceive a product also is not surprising (Table 2, part
2); until recently, even scholars of consumer behavior were unaware of construal level as a re-
search construct. For the more cognitive construct of certainty, a conscious reaction to a specific
situational stimulus is expected. The present research supports this view, namely, that consumers
deliberate about their certainty levels. However, the result of these deliberations are likely uncon-
scious, in that they might be aware they are willing to pay more for a product with a certain per-
formance level, but they probably do not realize that this increase in valuation stems from a
stronger elevation of the floor relative to the ceiling price of the WTP range.
The findings of this research offer further support for the feasibility of the S-O-R para-
digm in research into consumer behavior. The manipulations of the three situational variables
caused reaction within consumers: Consumers’ mood influenced information retrieval, stress af-
fected their level of construal, and certainty altered perceived risk of dissatisfaction with a prod-
uct choice. These consumer reactions all affected WTP, as the dependent behavioral response.
36
Table 2: Summary of manuscript results
37
The influence direction of the situational variables on consumers’ WTP often is not con-
stant but instead dependent on moderators specific to the situational variable. On the one hand,
the mood–WTP link is moderated by the type of product (mood-lifting vs. non–mood-lifting), in
combination with consumers’ culture (individualist vs. collectivist, Table 2, part 1; Table 3, stud-
ies 1.1 and 1.2). On the other hand, the characteristics and descriptions of the product under eval-
uation (high vs. low level of construal, Table 2, part 1; stress-reducing vs. non–stress-reducing,
Table 3, study 2) influence the relationship between consumers’ stress levels and their WTP. The
positive impact of consumer certainty on their valuation of products instead was established
without a moderator (e.g., no impact of position of consumers’ expected WTP relative to retail
price or of the valence of information about a product, Table 2, part 3).
Table 3: Summary of additional empirical investigation results
Although the impact of situational variables on consumer behavior might be expected
from extant research, a few results are surprising. The influence of between-subject differences in
regulating behavior, due to both mood and stress, on consumer WTP variance is novel and unex-
pected. Although previous research has established the possibility of mood regulation and stress
coping, it has not fully investigated the drivers of interpersonal behavioral differences. Manu-
script 1 (impact of mood on WTP) and empirical extension 2 (impact of stress on coping and
WTP) offer evidence that negative emotions (bad mood, stress) cause greater WTP variance for
38
products with regulation potential, driven by consumers’ different susceptibilities to regulating
behavior. Manuscript 1 also uses consumer culture as a means to operationalize the degree to
which consumers are prone to mood regulation. Therefore, this research contributes to debates
about mood regulation and coping with stress.
The individual manuscripts each offer unexpected insights as well. In the mood–WTP
relationship, the consistently mood-congruent behavior of more individualistic consumers was
surprising; in qualitative discussions, most consumers indicated they were prone to mood regula-
tion. However, this statement was neither theoretically supported nor substantiated by three main
(Manuscript 1) and two additional (empirical investigation 1) surveys. This research thereby adds
to the debate about the predominance of mood-regulating behavior.
Another surprising result was the positive impact of negative certainty (i.e., more reviews
with negative valence) on consumers’ WTP in study 1 of Manuscript 3. Intuitively, the more cer-
tain a consumer is about the bad performance of a product, the less he or she should be to pay for
it. Instead, the empirical results showed that consumers who saw more negative reviews were
willing to pay more, compared with those who viewed fewer negative reviews. This result is
counterintuitive and has not been explained sufficiently, such that it warrants additional research.
39
2. Implications
2.1 Research implications
Theory. The research presented herein yields relevant implications for three theoretical debates in
marketing research. First, for the discussion of consumers as homo economicus, the results of this
investigation imply that consumers are not fully rational beings with stable preferences and con-
stant, conscious utility maximization, in contrast with classic economic theories. This finding is
in line with many research findings in consumer behavior. Future theorizing and practitioners’
marketing mix choices thus should rest on the assumption that the consumer is not rational, and
the situation might interfere with consumers’ stable utility maximization. Instead, the same con-
sumer facing equal products at two different points in time is likely to behave differently, because
situational variables have interfered with his or her preferences.
Second, for research on consumer behavior, the present findings call for additional atten-
tion to situational variables, beyond consumer moods. Stress and certainty thus far have received
limited attention; they require additional research (Dost & Wilken, 2012; Mathur et al., 2006).
The empirical significance of both variables’ impact on consumer behavior supports this demand
for more attention. Multiple moderators of the mood–WTP link already have been identified (a
selection from more than 20 established moderators: product category, Gardner & Scott, 1990;
consumers’ openness to feelings, Chuang & Chang, 2007; consumers’ long-term well-being,
Zhong & Mitchell, 2012; gender, Thayer et al., 1994; social situations, Wang Erber & Erber,
2001), but fewer potential influences have been established for the relationship between stress
and WTP, referring to either construal level (e.g., product knowledge, Hong & Sternthal, 2010)
or coping (e.g., consumers’ social class, Henry, 2000). Still fewer refer to the certainty–WTP link
(e.g., risk aversion, Kahn & Sarin, 1988; affect, Laran & Tsiros, 2013). This list highlights the
discrepancy in research effort across fields, even though it is not complete, nor does the inclusion
of additional variables necessarily indicate better model quality (Coughlan et al., 2010). Theoreti-
cal approaches thus must recognize the influence of consumer stress and certainty on consumer
behavior. The relationships of stress and certainty with WTP also need further analysis to reveal
the respective links fully.
Third, this research supports a conceptualization of WTP as a range, informing the debate
about appropriate measures of consumers’ WTP. This range-based understanding appears to offer
40
both theoretical and empirical advantages. As the investigation of the impact of certainty on WTP
ranges shows, a point-based understanding of WTP might lag behind theoretical developments
(Manuscript 3, H2 and H3) and their empirical investigation. Specifically, changes in the size of
the WTP range would have gone unnoticed, and the effect of certainty on expected WTP incom-
pletely explained, if the dependent variable had been point based, because both these effects re-
quire an understanding of WTP as a range. Additional research that uses WTP as a dependent
variable therefore should measure the variable as a range. Even if the research hypotheses in
question predict only effects on the expected WTP, not on the floor or ceiling prices, an explora-
tory analysis of the WTP thresholds is meaningful. Only limited research thus far has adopted a
range-based conceptualization of WTP, so hypotheses based on extant research are unlikely to
yield implications for the WTP range.
On a related note, research on WTP measurement should extend its focus, from a method-
ological discussion (i.e., which measurement approach is most promising) to research into con-
sumer behavior more generally. In extant and the present research, the analysis has pertained only
to the impact of different certainty levels on the thresholds of the WTP ranges. However, it is
likely that multiple independent variables affect the components of the WTP range differently
(i.e., a variable causing a stronger shift in the floor relative to the ceiling price, or vice versa). The
self-limitation of research on WTP as a range on technical assessments of measurement tech-
niques thus forfeits the chance to contribute theoretically as well.
Methodology. The research efforts outlined in the enclosed manuscripts and additional investiga-
tions also yield methodological implications. First, when consumer behavior, particularly WTP,
is the dependent variable, situational variables need to be included as potential confounds. This
applies to both hypothetical and incentive-aligned WTP measurement techniques, as well as to
different degrees of participants’ reactivity, because situational variables likely do not interact
with strategic consumer behavior (e.g., the impact of consumers’ mood on WTP is unconscious,
so it should be similar irrespective of whether WTP has been stated hypothetically or not). Alt-
hough only the effects of consumers’ mood, stress, and certainty on WTP have been formally
established, other situational variables likely interact with them (e.g., sad mood and failing inter-
personal relationships lead to mood-congruent preferences, Lee et al., 2013; sad mood and an
option with a prevention focus increase selection likelihood, Baek & Reid, 2013) or have a direct
impact on consumer behavior (e.g., social visibility, Wysong et al., 2012; service encounter con-
41
text, Kim & Lee, 2012). If analyses of consumer behavior omit relevant situational variables,
measured variance could remain unexplained.
Second, the present research implies that several measures could decrease research costs.
Empirical investigations could use hypothetical measures, which are easier to generate, and sup-
plement them with incentive-aligned procedures. Incentive-aligned measures of WTP validated
the results generated by the hypothetical measures (e.g., studies 1.4 and 2.1), offering an early
indication that it is unnecessary to measure dependent variables only non-hypothetically. Howev-
er, none of the survey participants had any strategic interest in misstating their WTP, because the
research was branded as academic, not as a corporate pricing survey.
Another option to reduce research costs is online surveys, which offer an appropriate sub-
stitute for offline research. The results generated from online surveys (studies 1.3, 2.1, 3.2, and
3.3) matched the findings from the seven offline studies. Measures generated at the point of pur-
chase (studies 1.2 and 3.4) also were in line with those from experimental settings. Thus, the evi-
dence suggests that researchers may rely on online surveys, as long as external validity is con-
firmed by using alternative survey locations that can be better controlled (e.g., laboratory) or are
closer to the consumer (e.g., point of purchase). This advice might shift in retail environments for
which online shopping is a dominant purchase form (e.g., books and music in the United States),
because online surveys then would represent more realistic environments for testing behavioral
hypotheses.
Another pertinent measure entails the use of pretests, rather than manipulation checks,
which can reduce the main survey length and even potentially increase validity, by avoiding any
indications of an experimental manipulation (Herr et al., 2012). The studies of both stress and
certainty relied on extensive pretesting of the experimental manipulations, rather than manipula-
tion checks in the main study, and generated results in line with the hypotheses. Although not a
necessary indicator of the feasibility of pretests, this finding indicated their utility. Some
measures of manipulation strength also could not have been generated in a main study, such as
the physical assessment of participants’ arousal in the pretests for studies 2.2 and 2.3.
Third, the methodological implications of this study indicate the need for awareness of the
difficulties encountered, which could affect further research as well. Although the use of experi-
mental designs is common in consumer psychology and can help isolate certain situational influ-
ences on consumers, it suffers the risk of insufficient manipulation strength or incorrect manipu-
42
lation directions. Experimental conditions are comparatively easy to establish and control in a
laboratory setting, but their supervision is impossible in online experiments. This difficulty is
especially salient for online samples acquired through panels (e.g., Amazon MTurk), in which
participants often answer surveys professionally and thus might pay limited attention to the ma-
nipulations. For example, the additional empirical investigation of the effects of different review
distributions on consumer certainty and WTP ranges suffered from failed manipulations (III.3). If
experiments are to be conducted online, prestudies testing the manipulations also should be un-
dertaken online (e.g., study 3.1). An alternative would be to use non-experimental designs (e.g.,
measure the independent variable without manipulation, as in study 3.3), though a considerable
sample size is likely needed to ensure sufficient sample heterogeneity.
Fourth, researchers must be careful when designing experiments with stress or consumer
construal levels as variables, because stress is difficult to manipulate, and construal level is chal-
lenging to measure. In particular, low levels of stress are difficult to generate, because many par-
ticipants possess a generally high arousal level (e.g., due to daily hassles), which is difficult to
mitigate. An extensive manipulation to reduce arousal levels or a careful selection of situations
that cause consumers to exhibit lower arousal levels thus might be required. Finally, because con-
sumers’ level of construal is nearly impossible to measure (Pham et al., 2011), it might be easier
to prime.
2.1 Managerial implications
Three managerial implications go beyond the conclusions drawn from the manuscripts and the
research suggestions. First, any market research targeted at supporting pricing decisions should
acknowledge the impacts of consumers’ mood, stress, and certainty on their product valuation. If
situational influences on consumers’ WTP are disregarded in marketing research, the results are
likely to be distorted, and the resulting pricing decisions would lead to suboptimal results. Alt-
hough the situational determinants might be normally distributed among participants, external
conditions likely influence the sample overall (e.g., weather causes bad moods, Parker &
Tavassoli, 2000; prominence of stress after work, Hoel et al., 2001). Any measurement of WTP,
without either controlling or measuring the situational variables, thus is likely to be biased.
Second, this research has established situational variables as tools for influencing con-
sumer behavior, particularly for retailers. Most reactions of consumers to the three situational
43
variables are unconscious, so such stimuli appear especially powerful for marketers, in that the
consumer responses resulting from situational marketing stimuli tend to be non-strategic. From a
societal standpoint, the unconscious nature of the impact of the situational variables on consumer
behavior appears rather alarming; most consumers would not approve being subconsciously tam-
pered with by marketers hoping to make them pay more. Despite this normative objection
though, retailers might use situational variables to increase consumers’ WTP. Intuitively, happy
consumers spend more; the manuscripts detailing the impact of mood and stress on consumer
behavior also indicate ways to increase consumers’ valuation by encouraging them to enter a fa-
vorable state (good mood, low stress), such as through music (Oakes, 2003; Sar et al., 2011) or
pleasant smells (Morrison et al., 2011). Furthermore, a retailer might present peer preferences to
influence shoppers’ preference certainty (Narayan et al., 2011), and information at the point of
purchase could reassure consumers of the product’s future performance (Manuscript 3). Both
means to increase certainty likely improve consumers’ valuation of the products.
Third, products designed to help consumers manage their mood or stress levels might be
evaluated more favorably (Manuscript 1, empirical investigation 2). The package design of a
product with mood-lifting or stress-reducing possibilities should be adjusted to stress this poten-
tial and increase consumers’ valuation. However, the positive effect of a product’s potential regu-
lating capability is limited by moderating variables (e.g., consumer culture). Therefore, marketers
must evaluate if target consumers possess appropriate characteristics that leave them prone to
mood regulation (e.g., member of a collectivistic society) and adjust package designs or adver-
tisements accordingly (e.g., stress mood-lifting capabilities in collectivist societies).
44
3. Limitations and further research
“Indeed, it is a strange-disposed time:
But men may construe things after their fashion,
Clean from the purpose of the things themselves.”
—Cicero to Casca, Julius Caesar (I.iii)
Cicero saw through the traitor Casca’s description of various signs that he claimed announced
impending changes in the state of Rome (i.e., the assassination of Caesar; Shakespeare, 1957),
ascertaining Casca of his awareness that “men may construe things after their fashion.” Similarly,
consumer behavior researchers appear to see through the consumer. This investigation constitutes
no exemption: for instance, the impact of consumers’ stress levels on their WTP has been de-
scribed as mostly negative, explained via changes in consumer level of construal and character-
ized as dependent on the level of construal of the product under investigation (see manuscript 2).
However, the above quote issues two insights that challenge the predictive nature of the findings
of
Top Related